17936522. IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS simplified abstract (Capital One Services, LLC)
Contents
- 1 IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS
- 1.1 Organization Name
- 1.2 Inventor(s)
- 1.3 IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS - A simplified explanation of the abstract
- 1.4 Simplified Explanation
- 1.5 Potential Applications
- 1.6 Problems Solved
- 1.7 Benefits
- 1.8 Potential Commercial Applications
- 1.9 Possible Prior Art
- 1.10 Original Abstract Submitted
IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS
Organization Name
Inventor(s)
Phanindra Rao of Celina TX (US)
Chun-Hsiung Lu of McLean VA (US)
IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS - A simplified explanation of the abstract
This abstract first appeared for US patent application 17936522 titled 'IDENTIFYING PERFORMANCE DEGRADATION IN MACHINE LEARNING MODELS BASED ON COMPARISON OF ACTUAL AND PREDICTED RESULTS
Simplified Explanation
The patent application describes methods and systems for identifying performance degradation in machine learning models by comparing actual and predicted results. Here is a simplified explanation of the abstract:
- The system receives datasets of predicted and actual results for features in a system.
- It accesses a hierarchy of features and selects a level with a subset of features associated with a target feature.
- Impact values are generated for the subset, indicating contributions of features to the difference between predicted and actual results in the target feature.
- A new target feature with the highest impact value is selected from the subset, and a machine learning model is retrained for this new target feature.
Potential Applications
The technology described in this patent application could be applied in various fields such as finance, healthcare, marketing, and more to improve the performance of machine learning models by identifying and addressing performance degradation.
Problems Solved
This technology helps in detecting and addressing performance degradation in machine learning models, ensuring that the models remain accurate and reliable over time.
Benefits
The benefits of this technology include improved model performance, increased accuracy of predictions, and enhanced overall efficiency in utilizing machine learning models.
Potential Commercial Applications
A potential commercial application of this technology could be in the development of software tools for businesses that rely on machine learning models for decision-making processes. These tools could help companies maintain the accuracy and reliability of their models.
Possible Prior Art
One possible prior art for this technology could be the use of similar methods for identifying performance degradation in predictive models, but with a focus on specific industries or applications.
Unanswered Questions
How does this technology handle large datasets and complex feature hierarchies?
The patent application does not provide details on how the system manages large datasets or complex feature hierarchies when identifying performance degradation in machine learning models.
What types of machine learning algorithms are compatible with this system?
The patent application does not specify which types of machine learning algorithms can be used with this system for identifying performance degradation in models.
Original Abstract Submitted
Methods and systems are described herein for identifying performance degradation in machine learning models based on comparisons of actual and predicted results. The system may receive predicted and actual results datasets for features within a system with the predicted results being generated by a machine learning model corresponding to the feature. The system may access a hierarchy associated with the features and select a level of the hierarchy having a subset of features. The subset of features may be associated with a target feature. Impact values may then be generated for the subset, where the impact values indicate contributions of the corresponding features to a difference, in the target feature, between predicted and actual results. The system may select a new target feature, from the subset, associated with a highest impact value and may retrain a machine learning model associated with the new target feature.
